AN INTERNATIONAL PELLET ABLATION DATABASE

L.R. BAYLOR, A. GERAUD", W.A. HOULBERG, D. FRIGIONEb, M. GADEBERG", T.C. JERNIGAN, J. DE KLOEd, P. KUPSCHUS", B.V. KUTEEVe, P. LANGf, A.A.M. OOMENSd, A.L. QUALLS, K.N. SATOg, G.L. SCHMIDTh Oak Ridge National Laboratory, Oak Ridge, Tennessee, United States of America Association Euratom-CEA, Centre d'Qtudesnucleaires de Cadarache, St-Paul-Lez-Durance, France Associazione Euratom-ENEA, C.R.E. Frascati, Rome, Italy " J E T Joint Undertaking, Abingdon, Oxfordshire, United Kingdom Association Euratom-FOM, FOM Instituut voor Plasmafysica, Rijnhuizen, Nieuwegein, Net herlands e State Technical University, St. Petersburg, Russian Federation Max-Planck-Institut fur Plasmaphysik, Euratom Association, Garching, Germany g National Institute for Fusion Science, Nagoya, Japan Princeton Plasma Physics Laboratory, Princeton University, Princeton, New Jersey, United States of America
a

ABSTRACT. The contents are described of an international pellet ablation database (IPADBASE) that has been assembled to enable studies of pellet ablation theories that are used to describe the physics of an ablating fuel pellet in a tokamak plasma. The database represents an international effort to assemble data from several tokamaks of different magnetic configurations and auxiliary heating methods. In the initial configuration, data from JET, Tore Supra, DIII-D, FTU, TFTR, ASDEX Upgrade, JIPP T-IIU, RTP and T-10 have been included. The database contains details of measurements of deuterium and hydrogen pellet ablation, including pellet mass and speed, plasma electron density and temperature profiles, and pellet ablation light emission. A summary of the database contents and a scaling analysis of the data are presented.

1. INTRODUCTION
In order to predict the effect of fuelling the plasma with pellets in a next generation tokamak such as the International Thermonuclear Experimental Reactor (ITER), it is essential to have a pellet ablation model that adequately describes the experimental data in the current generation of devices. To date, most studies of experimental pellet ablation have concentrated on data from a single device [l-31 and therefore may not accurately reflect the general applicability of a particular theoretical model. An earlier effort to benchmark pellet ablation models with data from several different tokamaks was performed by Gouge et al. (41, but was limited by the quantity and quality of the ablation data. Another recent analysis by Kuteev [5]also examined a limited set of data from several devices. A more controlled set of experimental pellet ablation data is needed in which the pellet speed and size are known accurately and the background plasma parameters (temperature and density profiles) are well measured. It is the purpose of this
NUCLEAR FUSION, Vol. 37, No. 4 (1997)

effort , originating from discussions during the IAEA Technical Meeting on Pellet Injection held in Naka in May 1993, to provide such a database of controlled pellet ablation experiments that can be used to validate an improved pellet ablation model and to provide the necessary confidence for extrapolation of the model to the next generation of fusion devices, such as ITER. The primary purpose of this paper is to present the database to the community, together with a statistical description of the data. The authors fully expect and encourage other researchers to perform detailed analysis of the data with codes that use sophisticated ablation physics models. The authors look forward to the use of this database in the determination of the best ablation models to use for future large scale fusion devices.
2. DATA SELECTION

Ideally one would like to construct a pellet ablation database with data that cover a wide region of

445

BAYLOR e al. t

FIG.1. Plasma central electron density and temperature at the time of pellet injection for the recorda i the IPADBASE, identified by tokamak n device.

Table I. Range of Experimental Parameters for Ohmic, NBI and ICRH Plasmas

pellet and plasma parameter space. For a given tokamak device, the plasma parameter space is limited owing to the capability of the plasma to withstand a pellet perturbation and also limited because of collineairties in the plasma density and temperature. Therefore, it is desirable to use data from various tokamak devices with different ranges of parameters to obtain a more extensive selection of data. The data selected for inclusion in the database are the logical pellet parameters of interest, such as

the pellet speed and mass. Additional pellet data, such as pellet shape and temperature, are included in the database for completeness. Plasma parameters that are likely to be involved in the ablation physics are also included and consist of plasma electron density and temperature profiles, magnetic configuration, and auxiliary heating details. All auxiliary heating methods have been considered, even ones that produce suprathermal electron populations, in particular, electron cyclotron resonance heating (ECRH) and

446

NUCLEAR FUSION, Vol. 97,No. 4 (1997)

AElTICLE

AN INTERNATIONAL PELLET ABLATION DATABASE

FIG.2. Pellet masa and speed for the pellet events i the IPADBASE, n identified by tokamak device.

lower hybrid (LH) current drive or heating scenarios. In these cases, the ablation process is dominated by the fast electron contribution and the data given in the database must be supplemented by data on the electron distribution functions in order to model the ablation process. The corresponding shots are given as a reference and interested persons are encouraged to contact the responsible physicist. The range of pellet injection experimental parameters in the database for the different tokamak devices is tabulated in Table I. The ranges of plasma central electron density and temperature of the plasma during the pellet injection events in the database are shown in Fig. 1. There is a significant spread among the different devices but the spread in data for each individual device is somewhat leas. The ranges of pellet mass and speed are shown in Fig. 2 for all the pellets in the database from each device. The Tore Supra data have the largest variation in pellet speed due to having both a centrifuge and two-stage light gas gun injector. The mass variation seen in the JET and DIII-D tokamak data is due to having multiple size barrels in the injectors on those devices. The pellet mass is not currently measured in the Axially Symmetric Divertor Experiment Upgrade (ASDEXU) and Frascati Tokamak Upgrade (FTU) devices, and therefore nominal pellet massea are used. In the FTU data, the pellet mass is estimated from the
NUCLEAR FUSION, V l 97,No. 4 (1997) o.

density perturbation in the ohmic plasma assuming a fuelling efficiency of 80%.

3. DESCRIPTION OF PARAMETERS
The ablation process in the plasma is ordinarily monitored with a photodiode that observes the light emitted by the ablating pellet (assumed to be p m dominately D, or Ha light). The penetration depth of the pellets is usually determined from time of flight measurements of the pellet speed and from duration of the D, emission intensity, assuming the pellets have a constant radial velocity through the plasma, which was verified on JET by soft X ray camera measurements [2].Pellets that penetrate beyond the magnetic a i are not included because of the xs larger uncertainty in the penetration depth caused by strongly reduced D, emission. For the high speed pellet injection experiments carried out in Tore Supra, the penetration depth and the Da emission evolution are directly measured using a high dynamic range charged coupled device (CCD) camera. A standard CCD camera was used to measure the penetration depth when the centrifugal pellet injector was used and the Da emission was measured, as on the other devim, by photodiodes that view the ablation light emission with a suitable D, filter. The mass of the pellets is in general measured by a microwave cavity

447

BAYLOR et al.

Table 11. Results of Regression Analysis of the IPADBASE and Comparison with NGS Scaling
~~

Device JET DIII-D Tore Supra ASDEX-U' All databases All ohmic NGS model
* Note:

C (const.)
0.031 0.150 0.146 0.208 0.079 0.076 0.079

T, exponent
-0.89 -0.60 -0.84 -0.41 -0.51 -0.45 -0.56

n e

m P

UP

exponent
-0.30 0.23 0.07 0.07 -0.03 0.08 -0.11

exponent
0.24 0.10 0.18 0.06 0.12 0.11 0.19

exponent
0.41 0.27 0.27 0.18 0.32 0.35 0.33

Corr. coeff., R

0.91 0.76 0.70 0.69 0.67 0.82

The ASDEX-U data do not include a measured pellet mass. scaling study, the sensitivity of the pellet ablation to the profile shapes is not taken into consideration and could lead to discrepancies between cases with different plasma conditions and different devices. It is possible to linearize the above equation by taking the natural logarithm of both sides, thus yielding an equation whose coefficients can be determined through a linear regression technique [7]. An analysis of this type has been carried out on the database for the data from each device by itself and for the entire database. Database entries that include auxiliary heating that yields significant numbers of suprathermal electrons (ECRH and LH) are excluded from this scaling study for reasons discussed earlier. Only those devices whose data have a significant variation in the penetration depth scaling parameters (JET, Tore Supra, DIII-D and ASDEX-U) are analysed using the regression technique on their data alone. The pellets injected into ohmic plasmas with no auxiliary heating are also analysed separately. The results of the regression analysis for the individual devices and the entire database are shown in Table 11. A number of models have been proposed for the physics of pellet ablation in a fusion plasma over the past 20 years (Ref. [8] and references therein). Among these different models, the neutral gas shielding (NGS) model of pellet ablation [9] has become the most widely adopted model. Despite the fact that it only considers the shielding of the neutral cloud expanding from the pellet surface, uses a simple 1-D ablation geometry and assumes monoenergetic incident electrons, it has predicted pellet penetration depths that are in good agreement with a number of experiments. With the NGS model it is possible to derive a simplified relationship between the erosion
NUCLEAR FUSION, Vol. 37, No. 4 (1997)

in the injection line. The Q of the cavity changes in proportion to the dielectric change introduced by the pellet and thus with proper calibration can yield an accurate measure of the mass [6]. The plasma electron density and temperature profiles are measured by Thomson scattering diagnostics or interferometers and electron cyclotron emission (ECE) diagnostics on the different devices. The pellet velocity is commonly determined by measuring the time of flight between signals from two light barriers or between signals from a light barrier and a microwave cavity in the injection line. For centrifugal injectors, the velocity is given by the rotation frequency of the centrifuge arbor. 4. SCALING OF PENETRATION DEPTH The IPADBASE covers a wide range of pellet and plasma parameters, and thus is well suited to a scaling analysis of the pellet penetration depth. For our scaling study we choose to examine the penetration depth dependence on pellet mass, pellet speed, plasma electron temperature and electron density, since these are the database parameters that are most likely to directly affect the ablation of the pellet. We assume the penetration depth scaling to have a form given by

X/a = CT,(keV)" n,(102' m-3)b

x m,(102' atomslc wp(m/s)d
where X is the penetration depth, a is the plasma , minor radius, T is the central electron temperature, ne is the central electron density, mp is the pellet mass, wp is the pellet velocity, C is a constant and the exponents are the scaling law dependences that will be derived. It should be noted that in this 448

AELTICLE

AN INTERNATIONAL PELLET ABLATION DATABASE

FIG. 3. Comparison of penetration depth with scaling using regression analysis for the IPADBASE database.

rate of the pellet and the pellet size and background plasma parameters [lo]. If one assumes pellet injection normal to the plasma from the outside midplane and linear profiles for the electron temperature and density, a scaling of the penetration depth is obtained that has the relationship X/a = cTe'5/9 nL-1/9 ri/9 vi/3

where r p is the equivalent sPherid pellet radius and C is a constant that includes the atomic mass number of the ablatant and the plasma. In the NGS model the pellet equivalent spherical radius is used as a parameter' whi1e in Our database the pellet mass in units of number of atoms is used since the shape Of the pellet is not in general 'Pherical*Since the number Of Particles is Proportional to the equivalent spherical radius cubed' the penetration depth d i n g Of the NGS model that We would like to make the comparison with is given by X/a = CT'-5/9 nL-1/9, ? 2 / : $13
where m, is the pellet mass, and central electron temperatures and densities are used. The multiple correlation coefficient R i Table I1 n measures the strength of the dependence of X/a on the four variables. R2 is the proportion of the m i ance in the data that is explained by the resion [7]. The regression analysis for the database
NUCLEAR FUSION, V l 97,No. 4 (1997) o.

shows remarkably good agreement with the penetration depth scaling for all the parameters of the NGS model except the plasma central electron density, which varies somewhat among the different devices. This may be because of the fact that the actual density profiles are quite different from the assumed linear shape in the NGS model and that there is substantial variation of the plasma profile shapes in the differentdevices. In any , the dependence of the penetration depth data on density is rather weak as it is in the NGS model. The regression analysis from the individual machines and overdl database shows a distinct dependence of penetration depth on pellet velocity. Thisdoes not agree with the originalneutral gas and plasma shielding (NGPS) model [lo], where there is no dependence of pe]let penetration depth on the pellet velocity and electron density, but does agree with a more recent NGPS model [31. The penetration depth scalings from the more recent NGPS model [3] and from the cloud charging model [5] are Very similar to that from the NGS "iel Shown. In Fig. 3, we plot the penetration depth from the reSiOn scaling against the actual Penetration depth for the data from the different devices. This shows the scatter in the scaling data from the different deviand shows that d l the devices follow the regression scaling about equally well.

449

BAYLOR e al. t

The J E T data in the database show the largest correlation coefficient and the regression analysis of these data agrees well with the scaling from the NGS model. This is not surprising since an earlier detailed analysis of these data [2] yielded good agreement with the NGS model. The DIII-D data yield the most unusual regression scaling of penetration depth and this may be due to the type of plasmas that are encountered, which are H mode plasmas with significant neutral beam injection power. 5. DATA FORMAT AND ACCESS The presently released database is identified as IPADBASE 1.0 to distinguish it from future releases. It is desirable for the present database to be extended with more data, in particular from devices which were not considered in this first version. In order to achieve this, the experimentalists concerned are invited to contact the authors. The data in the IPADBASE 1.0 are stored in an EXCEL spreadsheet file for each tokamak device. These data are then easily transferable to personal computers via floppy disk or internet electronic mail. The data files that make up the database are in fact available to interested parties over the internet via the World Wide Web from the following address: http://www.ornl.gov/fed/pellet/Ornlpell.html. 6. SUMMARY A database has been established for hydrogenic pellet ablation data from a number of operating tokamak devices that covers a wide range of plasma operating space and pellet parameters. A statistical analysis of the data has been presented and shows some features that are consistent with the NGS model of pellet ablation; however, much additional detailed comparison with ablation physics models is needed

to determine the best physics models to use in current and future devices. The data in this database are made available to the fusion research community for detailed analysis and extrapolation to next generation fusion devices, such as ITER.

ACKNOWLEDGEMENTS
The authors gratefully acknowledge useful discussions and assistance in obtaining the pellet ablation data used in this study from the experimental teams on each device. Support and encouragement from Drs. M. Chatelier (CEA) and S.L. Milora (ORNL) are greatly appreciated. This research was sponsored in part by the Office of Fusion Energy Science, USDOE, under Contract No. DEAC05-960R22464 with Lockheed Martin Energy Research Corporation.

REFERENCES
BUCHL, K., et al., Nucl. Fusion 27 (1987) 1939. HOULBERG, W.A., et al., Nucl. Fusion 32 (1992) 1951. PEGOUFUE, B., et al., Nucl. Fusion 33 (1993) 591. GOUGE, M.J., et al., Fusion Technol. 19 (1991) 95. KUTEEV, B.V., Nucl. Fusion 35 (1995) 431. JENSEN, P.B., ANDERSEN, V., J. Phys., D, Appl. Phys. 15 (1982) 785. BEVINGTON, P.R., Data Reduction and Error Analysis, McGraw-Hill, New York (1969). MILORA, S.L., et al., Nucl. Fusion 35 (1995) 657. PARKS, P.B., TURNBULL, R.J., Phys. Fluids 21 (1978) 1735. HOULBERG, W.A., et al., Nucl. Fusion 28 (1988) 595.
\

(Manuscript received 5 March 1996 Final manuscript accepted 6 December 1996) (E-mail address of L.R. Baylor: baylorlr@ornl.gov)

450

NUCLEAR FUSION, Vol. 37, No. 4 (1997)